Various machines such as excavators, scrapers, articulated trucks and other types of heavy equipment are used to perform a variety of tasks. Some of these tasks involve moving large, awkward, and heavy loads in a small environment. Because of the size of the machines and/or the poor visibility provided to operators of the machines, these tasks can be difficult to complete safely and effectively. For this and other reasons, some of these machines are equipped with image processing systems that provide views of the machines' environments to their operators.
Such image processing systems assist the operators of the machines by increasing visibility, and may be beneficial in situations where the operators' fields of view are obstructed by portions of the machines or other obstacles. Conventional image processing systems include multiple cameras that capture different areas of a machine's environment. The cameras are placed strategically around the machines based on the machines visibility requirements. Individual live video streams from the multiple cameras may then be stitched together to form a partial or complete surround-view of the environment around the machines. Typically, the individual live video streams are stitched in a specific configuration with respect to each unique machine type and model. A single stitching configuration, however, is insufficient in articulated machines having several reference frames. When a single stitching configuration is used while articulated machines move, the changes in the cameras positions may distort the surround-view and even create a blind spot in the surround-view image.
One attempt to account for changes in the positions of the cameras in an image processing system is disclosed in U.S. Patent Publication No. 2014/0184799 (the '799 publication) to Kussel. The system of the '799 publication includes inclination sensors to determine changes from the initial orientations of the cameras. The system of the '799 publication also includes means for using information from the inclination sensors to correct deviations in the cameras' alignment over the life of the image processing system.
While the image processing system of the '799 publication may be used to stitch image data from cameras mounted on articulated machines, it requires separate calculations for each change detected in the inclination sensors. Consequently, huge amount of computations are necessary to account for movement of multiple cameras, as happens almost every time an articulated machine works. This is why the image processing system of the '799 publication fails to provide a real-time surround-view image using an image stitching process that dynamically changes to account for the changes in the cameras' relative positions.
The disclosed methods and systems are directed to solve one or more of the problems set forth above and/or other problems of the prior art.